A puzzle with three solutions

I wanted to briefly summarize the main results from our 718 work as we have it so far. This is a 3 species, 3 fleets, 5x5 map implementation of POSEIDON (data-limited). There are two main features that makes 718 special:

  1. Atrobucca brevis sudden price jump sometime in the recent past and its consequent change in SPR from close to 100% to less than 20%
  2. Lutjanus malabaricus consistently low SPR for at least the past 3-4 years

The second point represents the puzzle of 718. On the one hand, the SPR is consistently low and in fact lower than it was in 712. On the other hand L. malabaricus still is the major species landed in the area for all three fleets (in fact it is the only species of note caught by gillnetters). The overfishing implied by the SPR has not affected CPUE nor we know of any drop in effort or participation. More importantly, 718 is dominated by large boats with very high marginal costs, enticing large long-liners to travel back and forth from Probolinggo. Some of this can be explained by the earnings due to A. brevis, but even so the main species landed remain L. malabaricus.

In 718 we have used POSEIDON in “reverse” from other use cases like 712 or US West Coast. While there we took the data to produce a scenario on which to run simulations, here we produce many scenarios that fit the data and use the scenarios to understand what “hidden parameters” would cause the observations we see.
In the end I think the puzzle can be explained in only three ways:

  1. The SPR is wrong and there is no current overfishing. This is possible if we assume \(\frac M K\) is close to the global table rather than TNC estimates
  2. The SPR is right but the stock has other characteristics that make it extremely robust anyway (chiefly very high carrying capacity)
  3. The SPR is right but measuring local rather than total depletion and much more L. malabaricus remains untapped in the area

The first option is quite optimistic in the short run (L. malabaricus stock is not depleted) and requires only a light touch to steer L. malabaricus out of trouble. However it implies that marine protected areas are needed to rebuild A. brevis. This is what I presented already and is not the focus of this document. The second option is more pessimistic in the short run (L. malabaricus is depleted) but far more optimistic in terms of policy since the same features that kept the stock ticking until today will make even very light policies extremely effective. The third option requires no management of L. malabaricus, since it is still under-fished and will stay so for more than a decade; it will require a protected area for A. brevis.

Notice how all three scenarios are optimistic in a way 712 wasn’t. This is fundamentally due to the bio-economic nature of POSEIDON. All three fleets (local large longliners, long-range large longliners and large gillnetters) are expensive to operate and the total landings limit the amount of money available to pay for these boats. If the stock depletes to the point where these costs cannot be maintained, the boats will quit.
In a way it’s the opposite problem of what we had in 712. There boats are cheap to run so that the stock could reach far lower depths before becoming unstainable to exploit.

The third scenario is the least likely, in my opinion, because it can really only work if we assume that the fishery is still very young (7-8 years old on average) to generate the observations we see. The other two however really depend on where we fall on \(\frac M K\).

The thing about M/K

When you get down to it, we are trying to use the SPR measures to get an idea of the status of the stock. Embedding it into our agent-based model solves two of its weaknesses: (1) It is an equilibrium metric of mortality (where things would be if the same conditions applied repeatedly) and (2) needs to be weighted with other conflicting social information.
However there is a third issue that POSEIDON can’t really deal with and that is that SPR really is a function of the unknown parameters \(M\) and \(K\).
This is not an issue due to “formulas” or bypassable by some clever mathematical trick. It’s the main constraint of using length to inform status/mortality.

Basically, the lower the \(\frac M K\) ratio we assume, the more fish ought to be super-mature in a virgin population:

Its corollary is that the lower the \(\frac M K\) the more super-mature fish we need to sample in our catch in order to increase our SPR. In fact, unless L. malabaricus is caught almost exclusively above \(L_{\text{opt}}\) (dashed-blue line) there is little chance of hitting healthy SPR levels (assuming here M/K of 0.6.

So that’s eventually what the M/K debate comes down to. If we set M/K low, we are setting our task to be to bring the almost total landings to above 67 cm for L. malabaricus and nothing else will do.

We have already shown in a previous presentation what POSEIDON does when the prior for \(\frac M K\) is taken from FishLife (and so is centered around 1.2): it shows that the current landings are broadly sustainable with only minor policy adjustments. For the rest of the document I will then focus only on what happens if we force \(\frac M K\) to be below 0.9

Forcing POSEIDON to reckon with low \(\frac M K\)

Combining the data we have with low \(\frac M K\) implies that the fishery is exploited in an unstainable manner (low SPR) and has been so for a few years.
However in POSEIDON we are not conditioning just on SPRs. The filters include L. malabaricus being the main target species landed as well as the three fleets still being around:

Variable Target value
SPR A. brevis <10%
SPR L. malabaricus <10%
SPR L. laticaudis >40%
Landings A. brevis 600t - 2000t
Landings L. malabaricus 2500t - 5000t
Landings L. laticaudis 400t - 2000t
Gillnetters always active
Local longliners always active
Long-range longliners active after shock
2 year change SPR A. brevis drop by at least 40%
2 year change SPR L. malabaricus change by less than 10%

This means that we are looking at the corner of the parameter space where \(\frac M K\) is low (which here means below 0.9) but at the same time the fishery somehow having survived on L. malabaricus.
Besides the filters, the model also depends on the socio-economic data we collected which matters here because it assumes all three fleets to be expensive to run (and therefore requiring enough fish to make a profit).

The rejection filtering method tries to find a way for all these conditions to be true at the same time. It settles on a simple solution: if \(\frac M K\) is low but the fishery hasn’t collapsed then carrying capacity has to be higher. This is a bit of kludge but it means that, yes, the fishery is being overfished but fortunately it had a lot of stock to buffer it from a quick collapse. The upside of this is that the stock must then be very large and there are many potential gains from sustainable practices. The downside is that we have a long way to go to build it back.

The SPR of malabaricus is low, and the SPR of A. brevis has collapsed, like the data suggests. The total amount of biomass left for malabaricus is close to 10% but may have actually increased in the past 4 years. This is strange given that the SPR values are still very low. What happened is the boom in A. brevis prices combined with limits on how fast new entrants can join the fishery. As the price of A. brevis explodes many boats shifted from catching L. malabaricus to A. brevis. This respite has allowed L. malabaricus to recover slightly.

When no policy is implemented, what we have is the bust-phase of the A. brevis price boom: old-timers and new-entrants alike switch back from A. brevis to L. malabaricus, increasing its landings while pushing its biomass back down. Because A. brevis is still caught in its depleted state, the end result is that higher total earnings in the fishery will sustain a larger fleet in equilibrium, keeping both stocks low.

The equilibrium total earnings post-price boom are higher, sustaining a larger fleet. However, because of the importance of A. brevis and the higher price it commands for long-range boats, in the long run more and more effort will come from Probollingo based boats rather than locals (assuming price arbitrage is not corrected).

Policy in a low \(\frac M K\) world

Effort control

Because landings of L. malabaricus are bound to increase in the short run under business as usual, effort control tends to be less painful than in 712. Effort control involves postponing some of the landings increase in the short run for better stock health and higher landings five years down the line. Another way to look at it is to think that effort control mostly involves “locking in” the respite on the stock that the A. brevis boom brought.
On the other hand, since A. brevis is still expensive, effort control has only a limited effect on its depletion.

Marine Protected Areas

You need MPAs to have any effect on A. brevis. This is because with any effort control, it will be L. malabaricus landings that are responsive while A. brevis is just too expensive to avoid.
Here we implement a very mild MPA that covers only 5% to 10% of the total A.brevis habitat together with some effort control.
The immediate effects are very marginal on A. brevis but in the long run it adds up to a slight increase in the total biomass available.